The article introduces a novel computational approach called pkCSM (Predicting Small-Molecule Pharmacokinetic and Toxicity Properties Using Graph-Based Signatures) for predicting the pharmacokinetic and toxicity properties of small molecules. The method uses graph-based signatures to represent molecular structures and extract relevant patterns, which are then used to train machine learning models. The pkCSM platform includes 14 regression models and 16 classification models for various ADMET (absorption, distribution, metabolism, excretion, and toxicity) properties. The authors demonstrate that pkCSM performs as well or better than existing methods, with significant improvements in several data sets. A user-friendly web server is also provided to facilitate the rapid evaluation of pharmacokinetic and toxicity properties, without retaining any user-submitted information. The approach is scalable and can handle large data sets, making it a valuable tool for early-stage drug development.The article introduces a novel computational approach called pkCSM (Predicting Small-Molecule Pharmacokinetic and Toxicity Properties Using Graph-Based Signatures) for predicting the pharmacokinetic and toxicity properties of small molecules. The method uses graph-based signatures to represent molecular structures and extract relevant patterns, which are then used to train machine learning models. The pkCSM platform includes 14 regression models and 16 classification models for various ADMET (absorption, distribution, metabolism, excretion, and toxicity) properties. The authors demonstrate that pkCSM performs as well or better than existing methods, with significant improvements in several data sets. A user-friendly web server is also provided to facilitate the rapid evaluation of pharmacokinetic and toxicity properties, without retaining any user-submitted information. The approach is scalable and can handle large data sets, making it a valuable tool for early-stage drug development.